This is not true. If you "massively parallelize" an NP problem then you can solve it in polynomial time. That is exactly the definition of NP = Nondeterministic Polynomial time: If you can try all possible solutions at once and check which ones work, then the problem can be solved in polynomial time.
Yes, please. The idea that you can model "individuals", things like "how 'you' will respond to televised propaganda", and still expect to obtain any useful information to predict what will happen in the real world is *far* beyond our current levels in AI technology.
Modeling the world, to a level that can produce any useful predictions, is just damn too complex.
So the exit poll results are more accurate than the actual counted ballots. That's the premise of this book.
No, the premise of the book is that there are significant discrepancies between the two, which can not be easily explained as simple "accuracy errors".
I wish I had mod points. Most of the "interesting" messages in this thread are plain wrong and stupid.
If you had read TFA (I know, I know), you would knew that: * Alexander Ratushnyak was the only person to even meet the challenge's guidelines, because the guidelines for the contest are supposed to be hard to meet. * The contest is not about making a cool/fast/better general purpose compressor. It is a contest to see how good we can compress some particular body of text (a snapshot of Wikipedia), to investigate some interesting relations of compression with (artificial) intelligence. * The contest is not "closed" or just "finished", the contest is still open, anyone who thinks that can do better could send their copression tool for participation. As far as I remember, every time that someone can make better than the current record, he/she "wins the prize". * The posting about "I work a lot in Wikipedia and didn't knew about this, it should be low-profile" is just lame and stupid. This contest is trying to promote some very interesting area of research, and has nothing to do with Wikipedia other that they are using it as a body of text (human knowledge) to benchmark the tools and algorithms.
Please visit the site and read the related information. It is actually a very interesting research topic, and the relations with maths and artificial intelligence are also insightful.
I've actually learned about this contest and the theory from a previous story on slashdot. There are some nice mathematics and theoretical results that support this claim. This is also the motivation to have this 'contest', which for all people saying "20% is not much", "it takes too long to be used in applications"; well, this is not about creating better/faster applications to.zip your files, but more about testing the applicability of some interesting theoretical results.
Mod parent up. Indeed the game also seems broken for me.
Also parners in the ESP Game seemed less idiot. I have suggested about 20 labels for an image, and then I see: "Your partner has suggested 1 label"... WTF?!
Slashdot Mirror
I've got /usr/share/dict/words
$ grep '^[yuiophjklnm]\{12,\}$'
hypophyllium
miminypiminy
phyllophyllin
No, it doesn't.
God invented 0 and 1.
All the rest is the work of man.
This is not true. If you "massively parallelize" an NP problem then you can solve it in polynomial time. That is exactly the definition of NP = Nondeterministic Polynomial time: If you can try all possible solutions at once and check which ones work, then the problem can be solved in polynomial time.
Yes please, the 'logic' in GP post is all wrong.
MOD PARENT UP!
Yes, please. The idea that you can model "individuals", things like "how 'you' will respond to televised propaganda", and still expect to obtain any useful information to predict what will happen in the real world is *far* beyond our current levels in AI technology.
Modeling the world, to a level that can produce any useful predictions, is just damn too complex.
And why 3D? Just as an excuse to use the word Blogosphere?
but Möbius loops are more fun because you end up in sites that look like this!
I think they were the first to get it right: Reasonable quality of calls and very easy to setup on any sort of different configurations.
lol, please mod parent up
Mod parent up. Yes, this is the goal of the contest. Not making a faster/more efficient compressor.
I wish I had mod points. Most of the "interesting" messages in this thread are plain wrong and stupid.
If you had read TFA (I know, I know), you would knew that:
* Alexander Ratushnyak was the only person to even meet the challenge's guidelines, because the guidelines for the contest are supposed to be hard to meet.
* The contest is not about making a cool/fast/better general purpose compressor. It is a contest to see how good we can compress some particular body of text (a snapshot of Wikipedia), to investigate some interesting relations of compression with (artificial) intelligence.
* The contest is not "closed" or just "finished", the contest is still open, anyone who thinks that can do better could send their copression tool for participation. As far as I remember, every time that someone can make better than the current record, he/she "wins the prize".
* The posting about "I work a lot in Wikipedia and didn't knew about this, it should be low-profile" is just lame and stupid. This contest is trying to promote some very interesting area of research, and has nothing to do with Wikipedia other that they are using it as a body of text (human knowledge) to benchmark the tools and algorithms.
Please visit the site and read the related information. It is actually a very interesting research topic, and the relations with maths and artificial intelligence are also insightful.
I've actually learned about this contest and the theory from a previous story on slashdot. There are some nice mathematics and theoretical results that support this claim. This is also the motivation to have this 'contest', which for all people saying "20% is not much", "it takes too long to be used in applications"; well, this is not about creating better/faster applications to .zip your files, but more about testing the applicability of some interesting theoretical results.
. html
More information can be found here:
http://cs.fit.edu/~mmahoney/compression/rationale
yes, this is stupid.
Search on Google adding "site:wikipedia.org"
ModParentUp. I totally agree with this.
LOL. I wish I had mod points
Oh, no. Not again.
Nothing new to see here. Mice have long used humans to solve problems for them.
Mod parent up. Really nice video, and great ideas from this Luis Von Arn's. I enjoyed it A LOT!
Mod parent up. Indeed the game also seems broken for me.
Also parners in the ESP Game seemed less idiot. I have suggested about 20 labels for an image, and then I see: "Your partner has suggested 1 label"... WTF?!
How come that was modded as interesting?
Mod parent up. This idea is great for tagging images in the web, but it is not original from Google.